Employment Statistics and their abuse


In the past few weeks, I've seen several articles in right-wing media, such as the Wall Street Journal, that are "dissing" the Current Employment Statistics (Establishment survey) program at the Bureau of Labor Statistics, saying it's job counts are not in line with the "real world". And it's true that the Current Population Survey (Household survey) has been showing a rosier employment picture. Why is this such a big deal? Because the Republicans want to be able to say "Everything is OK", and if the numbers don't reflect that, well, get a new set of numbers, and discredit the inconvenient ones.

The Establishment survey covers nearly 400,000 establishments; the Household survey covers 60,000 households. All other things being equal, a sample of 400,000 yields better results than a smaller one.

According to this article, though, that discrepancy is only 140,000, not over a million, once you adjust for differing definitions and concepts of employment. Richard Berner at Morgan Stanley tries to get to the bottom of it:

United States: The Jobless Recovery and Monetary Policy

A key problem in figuring just how jobless this recovery has been lies in the data. The two leading measures of employment have diverged sharply over the 22 months of recovery, and the difference over the past year is still striking. The so-called payroll or establishment canvass shows a decline (not seasonally adjusted) of 560,000 nonfarm jobs over the past 12 months, and because that survey is derived from a sample of nearly 400,000 establishments, it is generally considered to be more reliable than the canvass of 60,000 households. In contrast, employment in the household survey rose by 842,000 from August 2002 to August 2003.

There are some intriguing twists to this gap that could mute the jobless recovery's impact on the economy's income-generating capacity. Some believe that those permanently laid off are now showing up as self-employed contractors, perhaps for the same firms, or that employment is growing, but it escapes the official payroll tally via the underground economy. But hard data to support those hunches is lacking (see "Is the Recovery Sustainable?" Global Economic Forum, September 8, 2003). And as Alan Krueger, a Princeton University economist points out, properly adjusting for the conceptual differences between the two measures shows that a modified household tally declined by 424,000 over the past year. That's because the payroll survey excludes workers in the self-employed, unpaid family, private household, and farm categories that are included in the household canvass, while excluding unpaid absences and including multiple jobs held. A technical adjustment connected with the 2000 census also boosted the household tally by 576,000 workers over the past year.

Now the payroll survey may still overstate job loss because it overlooks new establishments. So some upward revision to past payroll tallies wouldn't be a surprise, as often occurs in recovery. This week's report of the preliminary September job data will also contain preliminary tabulations of the "universe counts" for the first quarter of this year. They may indicate an upward revision for payrolls in March 2003, the benchmark month for extrapolating revisions. But the chance of a massive revision is small; the historical average for benchmark revisions over the last ten years has been plus or minus three-tenths of one percent.


Those rabid leftists at Morgan Stanley. Their analysis is sound, because they're in the business to look ahead, not to serve a political ideology.

Popular posts from this blog

New post-convention poll Results look good to me. Now, we have to keep them coming. Further down in the article, there is discussion of how Kerry is gaining ground as a person who understands the needs of people, and a person of character. Key problem is going to be keeping these issues in front of voters, when the Kerry campaign will have a finite amount of cash to campaign on. Bush won't have a finite amount until the Republican convention at the end of August. Other people are going to have to step up and start hammering the Bush administration. Kerry Leads Bush in Post-Convention Poll (washingtonpost.com) : The new poll shows Kerry now claims the support of 50 percent of all registered voters, compared with 44 percent for Bush, with independent candidate Ralph Nader at 2 percent. On the eve of the convention, Bush led Kerry 48 percent to 46 percent. Among those most likely to vote, the race is tighter: Kerry holds a 2-point advantage over Bush in the current po...
Read more
Sometimes headlines just write themselves Like this one from the Washington Post today: Ex-FEMA Head Starts Disaster Planning Firm Michael Brown certainly has shown he can plan a disaster. Look at how awful Hurricane Katrina was. For preventing or fixing things afterwards, he's not so good.
Read more
Reading List It is a good thing to be married to a librarian. Our local system has a web-accessable catalog where I can put a book on hold. Depending on how many people are waiting for the book, in a few days to a few months, my wife comes home with the book I've requested. I was reading "A Mathematician Plays the Stock Market" by John Allen Paulos. It is the kind of book that makes mathematics of investing accessable to non-mathematical types (and some of us mathematical folks as well), even as he tells the story of his failed love affair with Worldcom stock. Of course, I set this aside as soon as I got in "Big Lies: The Right-Wing Propaganda Machine and How It Distorts the Truth" by Joe Conason. This book takes all of the distortions of the right-wing media, the Bush-Cheney administration, and the Republican "conventional wisdom", and proceeds to refute them, citing facts chapter and verse. I think this book is great. But I've g...
Read more